MD5 performance

[Opencontent]

This pull-request boosts cluster performances (DB queries: durations) by having php (vs. postgres) calculating md5, together with some fixes (typos & phpdoc)

[francescor]

+1
this code has been already tested these days, but another series of test will be performed now. Any feed is really welcome.

[gggeek]

mmmm :-)

[gggeek]

Quite a lot of changes for a single pull request

[francescor]

many thanks, Gaetano!
should we split this pull request in smaller ones?

[andrerom]

@francescor yes, the PR says md5 being the topic, trying to solve everything and the kitchen sink in the same PR is not the way to go. Not easy to review and some of the fixes we might object to so then it delays the fix which is the important part. seems to be material for 4-5 PR's here; phpdoc, cs, md5, some return code changes, ...

[francescor]

ok, we'll do it..
in the meanwhile we've received good feeds from a pretty serious setup that have this code now: total runtime from 15s to 4s, so it's not peanuts ;)

[francescor]

(yes, we can confirm: cluster db queries duration from 33 secs to 1.5 secs)

[gggeek]

wow - seems like the postgresql query planner needs some serious help with optimizing away functions which have deterministic results... (either that or this is also an effect of the other changes you put in the PR).
A good tip for similar sql generated by the non-cluster php code, if any!

[francescor]

Gggeek, didn't get what you mean.. :(

[andrerom]

He is referring to applying the same optimization other places (legacy kernel for instance), not applicable to this PR so you can safely move on splitting this up so we can merge it :)